Spatial and temporal variation of dissolved heavy metals in the Lijiang River, China: implication of rainstorm on drinking water quality

Lijiang River is an essential drinking water source and natural scenery in the Guilin City. For the first time, implications of rainstorm were taken into consideration by investigating spatial and temporal variation of dissolved heavy metals (HMs) in the Lijiang River water. A total of 68 water samples were collected during low flow (normal) season and high flow (rainstorm) season from 34 sampling sites. Dissolved HMs including Cr, Mn, Co, Cu, Zn, As, Cd, Sb, and Pb were found to meet the respective drinking water standards, while comparatively higher concentration was observed after the rainstorm season, except for Cr. Multivariate statistical analysis showed that Co, Cu, Cr, Zn, Sb, and Pb in normal season were mainly controlled by anthropogenic sources. Furthermore, higher concentrations of Mn, Cu, Cd, Pb, Co, and Zn during the high flow season were attributed to rainstorm. The water quality index (WQI) showed good grades and comparatively lower in rainstorm season. The results of health risk assessment revealed that HMs in Lijiang River posed limited health risk; however, As posed potential health risk specially in rainstorm season. It is suggested to adopt preventive measures for mining activities and industrial waste-water discharge at the river’s upstream and downstream.


Introduction
The quality of water has significant impacts on human health and the ecological environment. Natural processes such as weathering, erosion, hydrological and climatic changes, volcanic activities, and landscape characteristics could be main factors of water quality deterioration (Ai et al. 2015;Meng et al. 2016). Besides, extensive technological developments coupled with a growing population have led to an increased demand for water, which in turn may bring about a decline in the quality and quantity of water resources globally. Anthropogenic impacts such as coal combustion, industrial or agricultural wastes, sewage discharge, mining, and transportation are among the manmade causes of river water pollution (Rakotondrabe et al. 2018;Villa-Achupallas et al. 2018;Wang et al. 2017). Taking into account the amount of freshwater per capita in China (80% sourced from the rivers) which is 1/3 of the global average, making it top priority issue to focus on. HMs in both dissolved and suspended phases are known as the most harmful types in aquatic environment among various types of pollutants (Shotyk et al. 2017). Although certain trace elements in limited concentration are indispensable for organisms, they pose threats to human or ecological security when safety thresholds are exceeded (Chowdhury et al. 2016). HMs such as As, Cd, Cr, Cu, Pb, Hg, Ni, and Zn which can cause cancer, Minamata, bone pain, and other illnesses are listed as the priority toxic pollutants (Strady et al. 2017). According to the health risk index, arsenic is considered the most significant pollutant for non-carcinogenic and carcinogenic problems and is a major contributor to chronic risk (Shahab et al. 2019(Shahab et al. , 2020Zeng et al. 2015). It was reported that about 6.8 million people in Bangladesh lived in high dosage of As (Chakraborti et al. 2010). The researches in Brazil also illustrated that Cr, Mn, and Ni pose non-carcinogenic risks (Ferreira et al. 2020). Therefore, HM level in drinking water resource should be clarified for the sake of public health safety.
The investigation of distribution, source identification, and health risk assessment of dissolved HMs in the rivers of China has been carried out previously (Wang et al. 2017;Xiao et al. 2019aXiao et al. , 2019bZhao et al. 2018). However, more attentions should be paid on the rivers in karst areas. Karst aquifers have special structures, geological backgrounds, hydrodynamic conditions, and hydro-chemical characteristics (Zhang et al. 2014), which may affect the migration and transformation of pollutants into the water. The Lijiang River is a typical karst river in which the bedrock is characterized by the pure and thick carbonate strata (Xu et al. 2016). Due to agricultural, industrial, and leisure activities, concentration of HMs in the sediments of Lijiang River are much higher than background and standard level Xu et al. 2016). Due to recent development trend in China, this relatively small watershed being one of the top tourist destinations is hosting nearly 100 million travelers annually (Shahab et al. 2020). Recent urbanization has caused environmental challenges such as road dust HMs sourced from fuel consumption (Shahab et al. 2020). These are on top of the usual industrial, residential, and agricultural contaminants that are daily entering the river without treatment.
Lijiang River is the main source of drinking water for the millions of citizens in the Guilin City . As rainy and humid monsoon is common climate in Guilin all the year, rainstorms could significantly affect the quality of Lijiang water. Rainstorm runoff is one of the key drivers on the rivers' water quality deterioration and carries and releases contaminants from the point and non-point sources to the consumable water bodies (Ma et al. 2015). Reports of HMs in Lijiang basin are alarming (Wu et al. 2020;Xiao et al. 2021;Xu et al. 2016); hence, it is essential particularly to investigate the influence of rainstorm in the study area. To the best of our knowledge, this is the first time to take the climate conditions into account in studying the dissolved HMs of Lijiang River water. In this study, a total of 68 water samples along the Lijiang River were collected systematically, to analyze nine dissolved HMs (i.e., Cr, Mn, Co, Cu, Zn, As, Cd, Sb, and Pb). The aims are (1) to investigate the spatial and temporal distribution characteristics and identify the sources of dissolved HMs in the Lijiang River water, (2) to evaluate the water quality and health risks of dissolved HMs in Lijiang River in different flow seasons through the water quality index and health risk assessment, and (3) to discuss the effects of rainstorm on the dissolved HMs and aquatic environment of the study area.

Study area
The Lijiang River (109°45′-110°56′ E, 24°38′-25°55′N) flows from the north toward the south of Guangxi province in southern part of China ( Figure 1). As one of the eight tributaries of the Pearl River (or the Zhujiang River), Lijiang River is approximately 164 km in length and originates from Maoer Mountains in hilly areas of Xing'an County that locates in the north of Guangxi province. The Lijiang River passes through the Guilin City from Lingchuan County to the Yangshuo County. At the downstream in Pingle County, it connects with Guijiang River and further joints to the Pearl River. The subtropical climate of the study area has monsoon-like weather. The average annual precipitation and temperature of the Lijiang basin is~1890.4 mm and~19.2°C (Li et al. 2017). Lijiang River is situated in the typical karst area characterized by the pure and thick carbonate strata (Xu et al. 2016). Karst aquifers have special structures, geological backgrounds, hydrodynamic conditions, and hydro-chemical characteristics (Zhang et al. 2014), which may bring higher risk of pollutants going into the water. Due to the increase of anthropogenic activities (tourism, mining, and industrial), HMs in Lijiang aquatic environment accumulated higher than the background values Xu et al. 2016). As the significant drinking water source for local resident, it is essential to pay more attentions on the water quality of Lijiang River.

Sample collection and analysis
A total of 68 water samples from 34 sites were collected in the upper reaches (S1-S9), middle reaches (S10-S25), and lower reaches (S26-S34) of the Lijiang River in two hydrologic periods. Sample collection was carried out during May (low flow season) and August (high flow season) after the rainstorm in 2019 ( Figure 1). For each site, 3 water samples from different depths (approximately at 0m, 0.5m, 1m) were collected with the water sampler and poured into a pre-cleaned polyethylene bottle and acidified with 1 mol/L HNO 3 .
Physicochemical parameters of water sample including pH, temperature, dissolved oxygen, and electrical conductivity were measured on-site with a multimeter. After sealing and labeling the bottles, the water samples were stored in a dark refrigerator at 4°C. Prior to measuring the concentrations of Cr, Mn, Co, Cu, Zn, As, Cd, Sb, and Pb, the water samples were filtered with 0.45-μm polycarbonate membrane and then analyzed by an inductively coupled plasma mass spectrometry (Nexion 350×; Perkin Elmer Ltd., USA). Standard materials (GSB 04-1767(GSB 04- -2004) and blank samples were set for quality control. The detection limits were ≤ 10 ng/L, and the recoveries ranged between 92.3 and 101.3% (Table S5).

Water quality index
The water quality index (WQI) is used to evaluate the quality of aquatic environment in this work. It reveals the impact of different parameters in the water samples. The WQI is calculated according to the equation below (Wang et al., 2017): where W i is the weight of water parameter i which displays the relative importance of parameter i among all the water parameters for drinking purposes; it is calculated according to the factor loading of parameter i and the eigen values from the PCA results (Table S1). C i is the measured concentration of HM i in samples. S i is the guideline value of HMi in the Chinese drinking water standards GB 5749-2006 (Tables 1 and 2) (Ministry of Health 2006). According to the value of WQI, the water quality of aquatic environment can be classified into five categories (Xiao et al. 2019a(Xiao et al. , 2019b, i.e., 0 ≤ WQI < 50 suggests the water quality is excellent; 50 ≤ WQI < 100 suggests is good; 100 ≤ WQI < 200 is poor; 200 ≤ WQI < 300 is very poor; and WQI ≥ 300 suggests the water is undrinkable.

Health risk assessment
The health risk assessment method was applied to quantify health risks (carcinogenic and non-carcinogenic) due to different exposure routes from HMs. In the health risk assessment model, the direct hand-mouth ingestion intake and dermal contact intake are usually considered as the two main exposure pathways for HMs in aquatic environment (Xiao et al. 2019a(Xiao et al. , 2019b. The exposure average daily dose for direct ingestion (ADD ing ) and dermal absorption (ADD der ) can be evaluated using the following equations (Chai et al. 2021): where C w is the mean concentration of each HM in the water sample (μg/L). The explanation and values of the IR, EF, ED, BW, AT and ET are showed in Table S2. The values of EF, ED, AT, and ET are referenced from US EPA (2004). The IR and BW are data of Chinese people cited from Zeng et al. (2015). K p is the dermal permeability coefficient of compound in water (cm/h, as presented in Table 3).
The hazard quotient (HQ) and hazard index (HI) show the potential non-carcinogenic risks of HMs. These methods are developed according to the individual's possibility, due to exposure to a certain dosage of contaminants (Alidadi et al. 2019). The HQ is, in fact, the ratio of the exposure dose (ADD) as compared to the reference, while HI is the accumulated HQs for each element. When the value of HQ or HI ≥ 1, the non-carcinogenic risk or adverse effects on human health should be considered; seriously, while for the values ≤ 1, the hazard is negligible. The HQ and HI are obtained using the below equations: where RfD is the reference dose (μg·kg −1 ·d −1 ); RfD ing and RfD der are shown in Table 3.

Multivariate and geostatistical analysis
In order to compare the spatial differences of 34 stations at different levels and identify the source of HMs, software including Origin2017, SPSS 19.0, and Microsoft Excel 2010 were applied to perform statistical analyses. The analysis of variance (ANOVA) was used to analyze the significant variation of HM concentration among different sampling sites. Pearson coefficients were determined by SPSS 19.0 which quantify the correlations among HMs. Principal component analysis (PCA) was adopted to investigate associations between the sampling sites and HM loads (Li et al. 2011;Ribeiro et al. 2018). ArcGIS 10.5 was employed to provide spatial distribution maps of HMs along the river. Inverse distance weighting method (IDW) was applied to estimate the value of non-sampling points by calculating the weighted average value of observed data from surrounding area (Teegavarapu and Chandramouli 2005), which is widely used in hydro-chemistry.

Concentration of dissolved HMs in Lijiang River water
The concentration of dissolved HMs in Lijiang River water during normal and rainstorm seasons are compared with drinking water standards (China, US EPA, and WHO) and listed in Table 1. It is reported that the higher coefficient of variation (CV) for HMs is affected by human activities, while lower CV revealed that HMs came from natural sources (Guo et al. 2012;Luo et al. 2021). The CV of dissolved HMs in normal and rainstorm seasons ranged from 23.46 to 320 and 25.87 to 117.8, respectively, showing that the distribution of HM concentration in Lijiang River water was quite variable, illustrating that the HMs in the Lijiang River could be perturbed anthropogenic factors. The ANOVA analysis results indicated that the concentrations of HMs except for As in normal and rainstorm seasons were significantly different (p < 0.05%, ANOVA), which revealed that rainstorm had an impact on the concentration of HMs in the water of Lijiang River. The average pH, temperature, dissolved oxygen, and electrical conductivity of Lijiang River water samples were 8.31, 26.0°C, 8.6 mg/L, and 117.57 μs/cm, respectively, and are within the standards of drinking water. The pH in the karst river region is usually high due higher carbonate content (Xu et al. 2020  Mn and Zn are consistently the most abundant metals in water samples during both periods, while Cd was the lowest. Although there were fluctuations in concentration of HMs; but were all within the drinking water standard limits in both normal and rainstorm seasons, which is in line with Yanghe River (Kuang et al. 2016). It is found that the concentration of Pb, Zn, Co, Cd, Mn, Sb, and Cu increased in rainstorm water samples, especially for Pb, Zn, and Co, which is 10 times higher than normal season. This may be related to the erosion of the surrounding soil by the river (Jiang et al. 2017) or weathering of the rocks (Xu et al. 2016). Furthermore, some factories in the vicinity also discharge industrial sewage into the river. However, Cr decreased in the rainstorm season. The exact reason is still unknown; however, previous studies reported lower concentration of Cr compared to other elements in the karst region because the alkaline environment (mean pH 8.31 in rainstorm season and 7.91 in normal season) enhances the precipitation of certain elements like Cr (Giri and Singh 2014;Xu et al. 2020).
According to Table S3, HMs in Lijiang River water were compared with other rivers. Comparison shows that concentration of Mn, Co, and Zn in Lijiang River is much higher than the world average background value, while Cr, Cu, and Cd values were in the similar level (Klavinš et al. 2000). Pb in Lijiang River water was lower than the world background value in normal season, however it increased significantly in the rainstorm season which is 7 times of the world background (Klavinš et al. 2000). Compared with other rivers in China, Cr, Cu, As, and Sb are generally lower in Lijiang River (Li and Zhang 2009;Wang et al. 2017;Wu et al. 2009;Zeng et al. 2019;Zeng et al. 2015), which suggests a better quality of Lijiang water. The Co concentration was similar in the Zhujiang River during low flow and high flow seasons (Zeng et al. 2019), while Co increased 10 times in the rainstorm season of Lijiang River. In general, concentration of Cr, Mn, Co, Cu, As, Cd, Sb, and Pb in the water of Catalan River (Carafa et al. 2011), Douro River estuary (Ribeiro et al. 2018), and Calore River (Zuzolo et al. 2017) is higher than the Lijiang River water. Although Zn is much lower in Catalan River, it keeps the similar level for the other three rivers.

Spatial and temporal distribution of dissolved HMs
Spatial and temporal distributions of dissolved HMs in Lijiang River are shown in Figure 2. According to the variation characteristics of Cr, Mn, Co, Cu, Zn, As, Cd, Sb, and Pb in the normal season of Lijiang River, the HMs were divided into three groups. Cr, Cu, Zn, Co, and Sb belong to the first group, which concentrations increased successively from upstream to downstream. The distribution of Cr, Cu, Zn, Co, and Sb in Lijiang River may be affected by human activities, which is similar to the Han River (Li and Zhang 2009). The second group were constituted by Mn, As, and Cd. High concentrations of Mn, As, and Cd were distributed uniformly in the upstream and middle reaches and lower in the downstream areas and are due to rock weathering and sediment dissolution which is common in the karst regions Xu et al. 2016). Pb is the only element in the third group with lower concentration in the upstream and downstream areas but higher in the middle reaches. The middle reaches of the Lijiang River lie in the city center regions where industrial and anthropogenic activities are common which probably has enhanced the Pb concentration (Shahab et al. 2020).
The categorization during the rainstorm season could be as follows. In the first group, distribution of Cr and Cd are more dispersed with high concentrations in the whole river. Mn, Cu, Sb, and Pb in the second group were the lowest at the upstream areas, while high concentrations were mainly distributed in the middle and down streams. It is similar to the results of normal season, which HMs may be influenced by dense human activities in the central areas (Shahab et al. 2020). The Figure 2 Spatial distributions of dissolved heavy metals in Lijiang River Basin. a Normal season; b rainstorm season third group includes Co, Zn, and As which the concentration decreased gradually from upstream to downstream.

Source identification of dissolved HMs
The correlation and interaction of 9 dissolved HMs in two flow seasons in the Lijiang River Basin via Pearson correlation matrix is shown in Table S4. In the normal season, Mn is significantly positively correlated with Co (0.658), Cu (0.483), and As (0.385). The concentration of Cr, Cu, and As is comparable with the background concentration suggesting its natural origin, from weathering of rocks and parent materials, and is transported to river (Xu et al. 2016). Cd exhibited significant positive correlations with Cu (0.429) and Co (0.368), and Pb (0.350) suggests natural weathering from sediments supplemented by anthropogenic pollution from fertilizers, traffic, and fossil fuel combustion which emit metal pollution like Pb, Zn, Cu, and Hg (Tang et al. 2010) and is transported by rainwater. Cr was significantly positively correlated with Zn (0.670), Sb (0.505), and Cu (0.446). Xu et al. (2016) reported that higher metals like Cu, Cd, and Zn are more bioavailable and have high exchangeable fraction from sediments to water. Pb was weakly positively correlated with Co (0.434), Cu (0.373), and Cd (0.389). The midstream of Lijiang River passes through the city downtown and is leisured by boating and cruise ship service. Combustion of fuel and transportation could possibly be the reason of these concentrated heavy metals (Shahab et al. 2020). In rainstorm season, Cr and Sb (0.503) and Cu and Mn (0.714) were positively correlated. Zn was positively correlated with Cd (0.656) and Co (0.511). There was a significant positive correlation between Cd and Pb (0.678), which may due to the weathering and release of Cd and Pb from sediments or rocks caused by the rainstorm. The Lijiang River is in a typical karst area, which contains high content of carbonate and organic matter , and HMs easily enter the water due to high speed of the river flow (Xu et al. 2016;Xu et al. 2020). As shown, weak correlation among other HMs suggests it comes from independent source.
The PCA of heavy metals in normal (N) and rainstorm (R) seasons resulted in three and four principal components (PCs), accounting for 69.85% and 75.84% of the total variance, respectively (PC, eigenvalue >1), as shown in Table 2 and Figure S1. Aligned with previous studies, conducted PCA of HMs in multiple water system changes was attributed to different variables (Rakotondrabe et al. 2018;Xiao et al. 2019aXiao et al. , 2019b. The principal components in the normal season (PC1 N , PC2 N , and PC3 N ) accounted the total variance of 37.46%, 20.68%, and 11.71%, respectively. The results showed that PC1 N had high load (> 0.75) on Co (0.820) and Cu (0.891), moderate load (0.75~0.5) of Mn (0.604) and As (0.698), and weak load (0.5~0.3) for Cd (0.454), which was similar with the correlation results. In addition, PC2 N was positively explained by the high load of Cr (0.746), Zn (0.693), and Sb (0.507), whereas PC3 N was positively explained by high load of Pb (0.754). In rainstorm season, the PC1 R , PC2 R , PC3 R , and PC4 R explained total variance of 27.71%, 21.60%, 15.18%, and 11.35%, respectively. PC1 R had high load on Cd (0.796) and Pb (0.797) but moderate load of Mn (0.683) and Cu (0.589). Furthermore, PC2 R was moderately loaded with Zn (0.698) and Co (0.648). Cr (0.673) and Sb (0.605) were moderately loaded by the PC3 R . PC4 R had weak loading on As (0.388).
Based on the grouping of PC1 N and PC1 R , the accumulation of Mn, Co, Cu, and Cd in Lijiang River water in normal and rainstorm season was mainly from natural erosion enhanced by the use of fertilizers. Lijiang River basin is a fertile land where agricultural activities and use of pesticides and fertilizers are common. Co and Cu in surface water were mainly due to natural weathering of the rocks and subsequent soil action (Meng et al. 2016). In general, Mn, As, and Cd in the environment may come from rock weathering and river washing (Bai et al. 2019). The presence of Pb in water may be attributed to the use of Pb-acid batteries, industrial emissions, and vehicle exhaust (Shahab et al. 2020). It suggested that the likely sources of Cr, Zn, and Sb of PC2 N were contributed by anthropogenic sources, such as agricultural and industrial activities. The concentrations of Zn and Co in rainstorm season were mainly concentrated in the upper and middle reaches. It indicated that PC2 R may be controlled by natural sources, which was correlated with impact of heavy rain on carbonate minerals (Xu et al. 2020). The high loading of Pb in PC3 N is attributed to anthropogenic sources and is also reported in Xiangjiang River (Zeng et al. 2015). The concentration of Sb increased obviously after the rainstorm in PC3 R , which may be due to the decomposition of Sb by carbonate rocks under the agitation of the rainstorm (Xu et al. 2020). Higher concentration of As (in PC4 R ) is reported by Shahab et al. (2020) in the karst study area region.

WQI and health risk assessment
The WQI values of water samples from different sites are shown in Figure 3. In the present study, WQI in normal and rainstorm seasons were ranged from 0.50~7.96 and 6.17~20.82, respectively. Therefore, Lijiang River water quality of both normal and rainstorm season can be categorized as good (WQI value less than 50) and suitable for drinking with later slightly higher than former. During the rainstorm season, the fast-moving current may wash the contaminated soil and riverbed into the water (Jiang et al. 2017). The increased turbulent flow can also re-suspend the settled contaminants and HMs from the sediment and discharge them to the river. In Figure 3, the WQI values of most upstream and midstream samples were higher than the average value in normal season, while in the rainstorm season, midstream and downstream samples were higher. It also demonstrated that heavy rains and scouring can affect river water quality, at least in terms of heavy metals. However, the WQI values of some sites in the midstream and downstream during normal season were lower than the average but in rainstorm season were above the average. This finding suggested that the human activities in the middle and lower reaches have a greater impact on water quality, which was similar with the results of geospatial interpolation. Therefore, we suggest that the decision-makers should pay attention to the prevention and control measures in these areas of the Lijiang River, including agricultural activities and industrial waste-water discharge. As shown in Table 3, the HQ ingestion , HQ dermal , and HIs values, for adults and children, of heavy metals (except As) are < 1, indicating that these HMs (oral ingestion and skin absorption) in the Lijiang River were all below the hazard levels and have very limited health   effects. The values of HQ ing , HQ der , and HIs in Lijiang River water increased during rainstorm reason, except for Cr, which indicated that the HMs in river due to heavy rainfall or human activities may lead to increasing health risks. As illustrated in Figure 4, the HQ ing , HQ der , and HIs values for children are higher than adults, which suggest that children suffer greater risk from the exposure of HMs. The proportion of health risks for HMs in Lijiang River water is HQ ing for children (54.49~66.836) > HQ ing for adults (27.079~33.168) > HQ der for children (0.006~12.039) > HQ der for adults (0.003~6.424). It shows that direct intake is the main exposure route. Furthermore, the HI values of adults and children in normal season were 0.315 and 0.635, respectively; the HI value of arsenic in adults and children was 0.953 and 1.92, respectively. Therefore, we conclude that As might have potential non-carcinogenic risk to human health, especially during rainstorm season. It also reported high risks associated with As in other rivers (Villa-Achupallas et al. 2018). Despite the potential risks, the results revealed that public health were not obviously affected by dissolved HMs in Lijiang River water in the current situation, agreed with the outcomes of WQI.

Conclusions
The concentration of heavy metals in normal season and rainstorm season in Lijiang River was evaluated from the aspects of spatial distribution, pollution degree, and health risk. It is concluded that concentrations of Cr, Mn, Co, Cu, Zn, Cd, Sb, and Pb varied between the normal season and the rainstorm season. Compared to the standard and other studies, the water quality of the Lijiang River was in good category with slightly higher values in the rainstorm season. The distribution trends of HM concentrations in the upper, middle, and lower reaches of the Lijiang River between normal season and rainstorm season are significantly different. The statistical analysis suggested that heavy metals like Co, Cu, Mn, and Sb do not greatly exceed the background value in normal season and naturally exist its concentration in rainstorm season that increases along with Zn, Cd, and Pb because the river scour the surrounding agricultural soil, which may increase the content of HMs in the river water, together with weathering of rocks or sediments under the hydraulic agitation. In general, the water quality was categorized as good-grade and can be used for drinking. Health risk assessment showed that the HMs were all lower than the hazard level, with limited health impacts (except for As). The Lijiang River water should be treated prior to considering drinking especially by children during the rainstorm season and needs to be checked by the local authorities.